[collectd] Bugfix and PR for Collectd running on an overloaded server

[Yves Mettier]

Hello,

I'm running Collectd on an overloaded server (with rrdcached 
configuration).
A vanilla Collectd will crash for several reasons :

* PR #640 : when new servers appear, there are many rrd files to 
created. With "CreateFilesAsync" set to true, lots of new threads appear 
(thread bomb) to create new rrd files because there are too many IO for 
what the server can do at the same time.

* Issue #661 : memory leak ! PR #640 will fix it.

* PR #690 : in some strage cases, the WriteQueuLimitLow/High parameters 
will not work well. This is a bugfix.

* PR #691 : the WriteQueuLimitLow/High feature (introduced with PR #280 
in collectd-5.4.0) is working well if you know what limits to set. But 
there is no way to know how big the Write Queue is (stats from the 
network plugin give information on the Read Queue). This PR adds 
statistics on the Write Queue and help to tune the 
WriteQueuLimitLow/High parameters.


I'm writing this mail mainly for Collectd Developers to ask them to 
integrate (or not) those patched as soon as possible because my Collectd 
server will soon be no more overloaded (which is the best fix for me). 
And I will not be able to test anything in my overloaded environment.

Regards,
Yves

-- 
- Homepage       - http://ymettier.free.fr                             -
- GPG key        - http://ymettier.free.fr/gpg.txt                     -
- C en action    - http://ymettier.free.fr/livres/C_en_action_ed2.html -
- Guide Survie C - http://www.pearson.fr/livre/?GCOI=27440100673730    -